• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.5
• /
• pp.169-176
• /
• 2020

Recently, deep neural networks (DNNs) have brought revolutions to many mobile and embedded devices by providing human-level machine intelligence for various applications. However, high inference accuracy of such DNNs comes at high computational costs, and, hence, there have been significant efforts to reduce computational overheads of DNNs either by compressing off-the-shelf models or by designing a new small footprint DNN architecture tailored to resource constrained devices. One notable recent paradigm in designing small footprint DNN models is sharing parameters in several layers. However, in previous approaches, the parameter-sharing techniques have been applied to large deep networks, such as ResNet, that are known to have high redundancy. In this paper, we propose a parameter-sharing method for already parameter-efficient small networks such as ShuffleNetV2. In our approach, small templates are combined with small layer-specific parameters to generate weights. Our experiment results on ImageNet and CIFAR100 datasets show that our approach can reduce the size of parameters by 15%-35% of ShuffleNetV2 while achieving smaller drops in accuracies compared to previous parameter-sharing and pruning approaches. We further show that the proposed approach is efficient in terms of latency and energy consumption on modern embedded devices.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.35S no.10
• /
• pp.144-155
• /
• 1998

The array weights that will maximize the SNR for any type of noise environment are determined by the function of the antenna design configuration and the directions of receiving target and interference signals. The conventional SLCs(sidelobe cancellers) using the SNR maximization perform worst from the saturation of the receiving system of main channel when the main antenna has pattern with high gain at the arrival angle of strong interference. In this paper, the new SLC is accomplished by using two independent antenna architecture. Main antenna is implemented with adaptive nulling, which is used for rejecting high-power interference primarily. Auxiliary antenna is realized with adaptive array for receiving interference signal to be suppressed completely, which has a characteristics of sufficient gain for every direction. The new SLC is implemented with above both antennas. We show that the new SLC, which consists of the adaptive nulling main antenna and the adaptive array auxiliary antenna, is useful in reducing the effect of strong interference like jammer, because the adaptive nulling at main antenna prevents its receiver and signal processor for saturation by strong interference. The proposed SLC has improved SNR over the conventional SLCs. The improved SNR at sidelobe region is typically more than 7 dB for a given test signal. Moreover, it improves the SNR of about 20 dB under strong interference at mainlobe.

• Journal of KIISE:Software and Applications
• /
• v.31 no.5
• /
• pp.595-604
• /
• 2004

Automatic analysis of concepts or semantic relations from text documents enables not only an efficient acquisition of relevant information, but also a comparison of documents in the concept level. We present a multiple cause model-based approach to text analysis, where latent topics are automatically extracted from document sets and similarity between documents is measured by semantic kernels constructed from the extracted topics. In our approach, a document is assumed to be generated by various combinations of underlying topics. A topic is defined by a set of words that are related to the same topic or cooccur frequently within a document. In a network representing a multiple-cause model, each topic is identified by a group of words having high connection weights from a latent node. In order to facilitate teaming and inferences in multiple-cause models, some approximation methods are required and we utilize an approximation by Helmholtz machines. In an experiment on TDT-2 data set, we extract sets of meaningful words where each set contains some theme-specific terms. Using semantic kernels constructed from latent topics extracted by multiple cause models, we also achieve significant improvements over the basic vector space model in terms of retrieval effectiveness.

• The KIPS Transactions:PartB
• /
• v.13B no.2 s.105
• /
• pp.77-82
• /
• 2006

Digital halftoning is the technique to obtain a bilevel-toned image from continuous-toned image. Among halftoning methods, the error diffusion method gives better subjective quality than other halftoning ones. But it also makes edges of objects blurred. To overcome the defect, we proposes the modified error diffusion to enhance the edges using the property that human vision perceives the local average luminance and doesn't perceive a little variation of the spatial variation. The proposed method computes a spatialactivity, which is the difference between a pixel luminance and the average of its $3{\times}3$ neighborhood pixels' Iuminance weighted according to the spatial positioning. The system also usesof edge enhancement (IEE), which is computed from the normalized spatial activitymultiplied by the average luminance. The IEE is added to the quantizer's input pixel and feeds into the halftoning quantizer. The quantizer produces the halftone image having the enhanced edge. The computer experimental results show that the proposed method produces clearer bilevel-toned images than conventional methodsand the edge of objects is preserved well. Also the performance of the preposed method is improved, compared with that of the conventional method by measuring the edge correlation and the local average accordance at some ranges of viewing distance.

• Journal of Korea Water Resources Association
• /
• v.43 no.8
• /
• pp.721-731
• /
• 2010

The rainfall-runoff relationship is very difficult to predict because it is complicate factor affected by many temporal and spatial parameters of the basin. In recent, models which is based on artificial intelligent such as neural network, genetic algorithm fuzzy etc., are frequently used to predict discharge while stochastic or deterministic or empirical models are used in the past. However, the discharge data which are generally used for prediction as training and validation set are often estimated from rating curve which has potential error in its estimation that makes a problem in reliability. Therefore, in this study, water stage is predicted from antecedent rainfall and water stage data for short term using three models of neural network which trained by error back propagation algorithm and optimized by genetic algorithm and training error back propagation after it is optimized by genetic algorithm respectively. As the result, the model optimized by Genetic Algorithm gives the best forecasting ability which is not much decreased as the forecasting time increase. Moreover, the models using stage data only as the input data give better results than the models using precipitation data with stage data.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.4
• /
• pp.353-359
• /
• 2021

The Army is constructing a training system using Miles equipment that applies the latest science and technology to carry out military training. The Miles training system is a system that uses Miles equipment to simulate the damage situation of combat personnel and equipment in the same way as an actual battlefield by conducting practiced maneuvers in the field. Through this, the training force can experience conditions similar to an actual battle. In particular, the training effects of the warriors participating in the training can be maximized by establishing an integrated system that utilizes cutting-edge science technologies, such as information communication and computer simulation. This study analyzed the effects of Miles training in the army using scientific techniques targeted at the mid-range Miles. In particular, the effect index for analyzing the training effect was derived from a literature survey and expert opinions. The weight of each effect index was calculated by applying the Swing method. The final training effect was calculated by combining the results of the survey from train-experienced people. The Miles training effect was 2.6 times more effective than previous training without using Miles, and the satisfaction rate with Miles training according to status was high through variance analysis, and the difference was statistically significant.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.3
• /
• pp.100-109
• /
• 2019

Maintainability indicates the extent to which maintenance can be done easily and quickly. The consideration of maintainability is crucial to reduce the operation and support costs of weapon systems, but if the maintainability is evaluated after the prototype production is done and necessitates design changes, it may increase the cost and delay the schedule. The evaluation should verify whether maintenance work can be performed, and support the designers in developing a design to improve maintainability. In previous studies, the maintainability index was calculated using the graph theory at the early design phase, but evaluation accuracy appeared to be limited. Analyzing the methods of evaluating the maintainability using fuzzy logic and 3D modeling indicate that the design of a system with good maintainability should be done in an integrated manner during the whole system life cycle. This paper proposes a method to evaluate maintainability using SysML-based modeling and simulation technique and fuzzy logic. The physical design structure with maintainability attributes was modeled using SysML 'bdd' diagram, and the maintainability was represented by an AHP matrix for maintainability attributes. We then calculated the maintainability using AHP-based weighting calculation and fuzzy logic through the use of SysML 'par' diagram that incorporated MATLAB. The proposed maintainability model can be managed efficiently and consistently, and the state of system design and maintainability can be analyzed quantitatively, thereby improving design by early identifying the items with low maintainability.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.1
• /
• pp.51-58
• /
• 2023

In this paper, we propose an organ segmentation technique for the automatic extraction of medical diagnostic indicators from X-ray images. In order to calculate diagnostic indicators of heart disease and spinal disease such as VHS(vertebral heart scale) and Cobb angle, it is necessary to accurately segment the thoracic spine, carina, and heart in a chest X-ray image. A deep neural network model in which the high-resolution representation of the image for each layer and the structure converted into a low-resolution feature map are connected in parallel was adopted. This structure enables the relative position information in the image to be effectively reflected in the segmentation process. It is shown that learning performance can be improved by combining the OCR module, in which pixel information and object information are mutually interacted in a multi-step process, and the channel attention module, which allows each channel of the network to be reflected as different weight values. In addition, a method of augmenting learning data is presented in order to provide robust performance against changes in the position, shape, and size of the subject in the X-ray image. The effectiveness of the proposed theory was evaluated through an experiment using 145 human chest X-ray images and 118 animal X-ray images.

• Journal of the Korean Society for information Management
• /
• v.40 no.1
• /
• pp.121-148
• /
• 2023

The purpose of this study is to propose a scholarly paper recommendation system based on metadata attribute similarity with excellent performance. This study suggests a scholarly paper recommendation method that combines techniques from two sub-fields of Library and Information Science, namely metadata use in Information Organization and co-citation analysis, author bibliographic coupling, co-occurrence frequency, and cosine similarity in Bibliometrics. To conduct experiments, a total of 9,643 paper metadata related to "inequality" and "divide" were collected and refined to derive relative coordinate values between author, keyword, and title attributes using cosine similarity. The study then conducted experiments to select weight conditions and dimension numbers that resulted in a good performance. The results were presented and evaluated by users, and based on this, the study conducted discussions centered on the research questions through reference node and recommendation combination characteristic analysis, conjoint analysis, and results from comparative analysis. Overall, the study showed that the performance was excellent when author-related attributes were used alone or in combination with title-related attributes. If the technique proposed in this study is utilized and a wide range of samples are secured, it could help improve the performance of recommendation techniques not only in the field of literature recommendation in information services but also in various other fields in society.

• Journal of Animal Science and Technology
• /
• v.55 no.1
• /
• pp.7-11
• /
• 2013

The purpose of this study was to estimate (co) variance components of three milk production traits for genetic evaluation using a multiple lactation model. Each of the first five lactations was treated as different traits. For the parameter estimation study, a data set was set up including lactations from cows calved from 2001 to 2009. The total number of raw lactation records in first to fifth parities reached 1,416,589. At least 10 cows were required for each contemporary group, herd-year-season effect. Sires with fewer than 10 daughters were discarded. Lactations with 305d milk yield exceeding 15,000 kg were removed. In total, 1,456 sires of cows were remained after all the selection steps. A complete pedigree consisting of 292,382 records was used for the study. A sire model containing herd-year-season, caving age, and sire additive genetic effects was applied to the selected lactation data and pedigree for estimating (co) variance components via VCE. Heritabilities and genetic or residual correlations were then derived from the (co) variance estimates using R package. Genetic correlations between lactations ranged from 0.76 to 0.98 for milk yield, 0.79~1.00 for fat yield, 0.75~1.00 for protein yield. On individual lactation basis, relatively low heritability values were obtained 0.14~0.23, 0.13~0.20 and 0.14~0.19 for milk, fat, and protein yields, respectively. For the combined lactation heritability values were 0.29, 0.28, and 0.26 for milk, fat, and protein yields. The estimated parameters will be used in national genetic evaluations for production traits.